eGFR is an Estimate.

There is nothing in medicine that wastes more of my time than the reporting of the eGFR. I never go a week without a patient calling me, worried that he or she is headed for dialysis, because of the eGFR reported to them on MyChart. Beyond my trifling complaints, there have been serious discussions about the race-based adjustments to GFR.

Here, Dr. Eric Cohen explains an even more systemic problem with the eGFR. The post is concise, and although it may seem important only to healthcare workers, I suggest it is worth everyone’s time. It is also a great pre-read for tomorrow’s Fortnight Podcast (how is that for foreshadowing).

Adam Cifu

Measuring kidney function is important in health and disease. Its estimate as eGFR is now routinely reported by clinical labs, calculated from the values of serum creatinine, age, and gender. Most practitioners refer to the reported values as “GFR”, that is, glomerular filtration rate, and omit the “e”. But the “e” stands for estimate, not electronic, not “equal to”, but estimate. Indeed, large studies show that eGFR does not closely correspond to measured GFR. The NIH states that “For even the most accurate estimating equations, only about 90% of GFR estimates are within 30% of the measured GFR of the study population (P30), with varying accuracy across populations.”

Within 30%. So, we accept an estimate that is within 30% of correct. That is remarkable, especially when that number is bandied about without its “e” prefix. Worse still, often the eGFR is reported with a decimal precision, for instance, 46.7 ml/min instead of 47 ml/min (per 1.73 m²).

But the eGFR may be even more misleading than its up-to-30% inaccuracy. Its value is almost always reported as being indexed to a body surface area of 1.73 m². That 1.73 is a relic from one hundred years ago, when the average body surface area (BSA) in the USA was 1.73 m². The average BSA today is near 1.83, ~ 1.91 in men and 1.71 in women. Thus, for any printout showing eGFR in an American male, one can assume that the true eGFR is almost ten percent higher. And for a smaller woman, the reported eGFR could be 5% lower than the true GFR.

De-indexing the eGFR to the patient’s actual BSA is probably wise, and some are starting to do this, especially for drug dosing. After all, one wants to have an estimate of the GFR that is specific to the individual, rather than one normalized to fit the needs of the epidemiologist.

Imprecision becomes nonsense when the “eGFR” is reported for patients with acute kidney injury, who are not in the steady state as far as their serum creatinine is concerned. The eGFR in such patients is meaningless because the serum creatinine value lags behind any change, good or bad, in the GFR. The many studies that have created equations for estimating the GFR have been done in patients who are in a steady state with stable renal function. Yet eGFR is reported in subjects with acute renal failure and used by some clinicians as a metric assumed to be better than the simple serum creatinine.

Then there’s the impossibility of eGFR in chronic dialysis patients, who have no, or very little, GFR of their own.

Suggestions and recommendations:

1. The eGFR should never be reported to decimal precision.

2. Involved clinicians should be periodically reminded that the “e” in eGFR stands for estimated.

3. Leaders in Clinical Chemistry should strongly consider to modernize the index BSA to 1.83 m².

4. eGFR must never be reported for patients on dialysis.

5. eGFR should not be reported for patients in acute renal failure.

Eric P. Cohen, MD, is a Nephrologist at the Manhattan VAMC and New York University School of Medicine.

Photo Credit: Volodymyr Hryshchenko